Espace membre

Cet espace est dédié aux étudiants, aux enseignants et au personnel administratif de l'école


Mot de passe oublié?

Actualités de l'école


Doctorate thesis defense of Asma Ben Hadj Hmida

Doctorate thesis defense on April 10st 2017 at 14H00 AM ,in Amphi I, Sup’Com.

Entitled :Multiple Antenna Spectrum Sensing Techniques for Cognitive Radio Applications

Presented by : Asma Ben Hadj Hmida 




Professor at SUP’COM, Tunisia




Reviewers :

Mrs. Noura SELLAMI

Professor at ENIS, Tunisia


Mr. Ridha HAMILA

Professsor at Qatar University, Qatar



Mrs. Leïla NAJJAR

Associate professor at SUP’COM, Tunisia


Supervisor :

Mr. Sofiane CHERIF

Professor at SUP’COM, Tunisia


Co-Supervisor :

Mr. Hichem BESBES

Professor at SUP'COM, Tunisia



Cognitive Radio (CR) has appeared, in the recent years, to deal with the leakage of spectrum resources via exploiting, opportunistically, the unoccupied frequency bands of the licensed user by an unlicensed one. To do so, spectrum sensing should be performed in order to detect the free bands. A key challenge is to establish an efficient and reliable detector that guarantees a tradeoff between the low computational complexity, the detector design simplicity and the robustness to an uncertainty on the estimation of a certain required information, such as the noise level or the transmission channel gain.

This thesis contributes in the field of the standalone spectrum sensing using the spatial cross-correlation of the antenna's output, denoted by CSOM, in order to deal with the robustness to noise estimation uncertainty drawback of the classical Energy Detector while maintaining its design simplicity. Three main sensing techniques are drawn based on the choice of the weighting vector of the CSOMs and the considered application.

On the one hand, an adaptive approach, given the channel gain's cross-correlation coefficient, is proposed in order to optimize the sensing parameters through maximization of the NDC. Performance analysis in terms of spectral efficiency and decision reliability shows enhancement of the proposed approach compared to classical ones.

Therewith, a highlight on the effect of the number of the considered CSOMs on the decision is revealed. Then, a combination between the NDC and the MINLP with binary constraint for the weighting vector, is proposed in order to optimally select the moments. Simulations reveal the impact of the algorithm via elimination of the weak moments from decision, which lead to the increase of the spectral efficiency while decreasing the sensing delay.

On the other hand, for the blind sensing, we contribute on the development of two main detectors that harness the statistical characteristics, afforded by two techniques, namely the kurtosis and the Goodness-of-fit test of the CSOM. Empirical and analytical performance analysis are exhibited.

Besides, for a simpler detector design, it is substantial to derive the analytic performance of the different presented sensing approaches. To this end, statistical characterization, namely the pdf of the inner product of two dependent complex random Gaussian vectors is mandatory in this work either for the resolution of the optimization problems or the analytical threshold determination. However, to the best of our knowledge, no previous studies have attended to address, thoroughly, this problem, which motivates us to the second elementary contribution of this thesis. This can be also substantial for other signal processing application, such as radar, multiplicative noise and channel estimation based applications.


Cognitive Radio; Spectrum Sensing; Probability density function; Receiver Operating Characteristics (ROC)